7/Dive into Superbuilders in our latest piece ⬇️ .

6/ Superbuilders also enable MEV capture across chains, estimated to be worth $3.6B by @flashbotsmev as well as allow apps to access liquidity from all chains in the network and giving consumers the ability to keep their funds on one chain.

5/ Superbuilders address this concern by guaranteeing atomic execution. They read the state of each rollup and can build across all of them, allowing users to interact across multiple chains as seamlessly as they can two apps on the same chain today!.

4/ Shared sequencers were created to solve this problem by unifying chains through their sequencer, but they could only get so far. Why? They can only guarantee that transactions will be included (atomic inclusion), but not that they will actually execute.

3/Composability would allow any user to transact with funds on any chain without bridging and would enable applications to access liquidity across all chains in the network. The problem now: each new rollup produces its own block of transactions, keeping them siloed.

2/Congestion of the Ethereum L1 led to the creation of rollups to expand the capabilities of Ethereum. While scaling allowed for cheaper costs and more transactions, it did so at the cost of composability between all of the users and applications.

Shared Sequencers are Dead, Long Live Superbuilders. Scaling blockchains like Ethereum with L2s sacrifice an important element of blockchains: composability. Shared sequencers sought to restore this, but haven’t gotten all the way there. Enter Superbuilders.

Interested in how this technology can also secure Machine Learning networks? . Learn more in our deep dive on the verification problem:.

5/This method minimizes verification costs by focusing only on disputed computations, making decentralized networks faster and more trustworthy.

4/That specific computation is executed on-chain. Blockchain nodes each run the computation and consensus is determined. The party in the wrong gets penalized and has their stake redistributed.

3/They use a binary search on the execution trace. By repeatedly splitting the trace in half, the exact step that causes the discrepancy is found.

2/The challenger and sequencer each create an execution trace-- a detailed record of each computational step and its resulting state.

1/A sequencer posts the claimed state of its L2. During the challenge period, a challenger, having run the same transactions and believing the sequencer is incorrect, kicks off the dispute process.

Optimistic Games Explained in 4 Easy Steps: A Thread. Ever wonder how an optimistic rollup works? Here's a good mental model

Decentralized compute networks are the future, but they face a critical hurdle: the verification problem. How do you know the output you get is what you asked for? Short answer: you don't. Our deep dive into the verification problem for inference🔽.

6/ The good news? Innovation is underway. Learn more about this technology and its implications in a recent Coindesk article written by our team!⬇️. .

5/ ZKPs have a long way to go. Right now, they're 4-6 orders of magnitude slower than running computations natively. To put that in perspective: high end systems are proving computations 150x slower than running them on a TI-84 calculator. 🤯.

4/ 👀 Teams like @gensynai and @InferenceLabs are working to deploy decentralized training and inference using consumer hardware. The challenge? Tough engineering problems, like the floating point problem.

3/ 💡 The tech behind ZKPs could be a game-changer for decentralized systems. Idle compute power on your MacBook could help build models and earn you rewards.

2/ ZKP allows one party (the prover) to convince another (the verifier) that a statement is true—without revealing the statement itself. The cool part? Other nodes can quickly verify this proof and confirm everyone’s playing by the rules. 🔐.